Economic Consequences of Federal Farm Commodity Programs, 1953-72

ECONOMIC CONSEQUENCES OF FEDERAL FARM COMMODITY PROGRAMS, 1953-72

By Frederick J. Nelson and Willard W. Cochrane*

Farm programs of the Federal Government kept farm prices and agriculture, (2) the productivity of measured agricul- incomes higher than they otherwise would have been in 1953-65, tural resources, and (3) agricultural output levels. Such thereby providing economic incentives to growth in output sufficient to keep farm prices lower than otherwise during 1968-72. long-term changes did not occur independently of the The latter result differs significantly from findings in other farm programs. These programs were operated in a way historical free market studies. These conclusions stem from an that reduced risk and uncertainty for farmers, affected analysis of the programs in which a two-sector (crops and live- their expectations of future income potential from stock) econometric model was used to simulate historical and farm production activities, and influenced their willing- free-market production, price, and resource adjustments in U.S. agriculture. Supplies are affected by risk and uncertainty in the ness and ability to invest, to adopt cost-reducing tech- model, and farm technological change is endogenous. nology, and to adjust output levels. Keywords: Government farm programs, farm income, risk, In considering effects of the programs, it is desirable technological change, free market. to specify a model in which shortrun and longrun agricultural output responses are affected by invest- THE OBJECTIVE ments, current input expenditures, and farm technological changes. These, in turn, should be influenced by price and income expectations and experiences, by the Policy decisions affecting future production, con- extent of risk and uncertainty, and by technological sumption, and prices of food and fiber in the United change. Such ideas were used in developing this model. States need to be made with as full knowledge as possi- A unique feature of the model is that it includes endoge- ble of the likely longrun and shortrun consequences. The nous risk and resource productivity proxy variables. quantitative analysis of past farm commodity programs Not much quantitative knowledge exists about described here can provide useful information for intermediate and longrun supply adjustments under a analyzing the consequences of future alternative sustained free-market situation. No claim is made programs. however, that this model's results represent the defini- How would agricultural economic development in tive word in free-market analysis of the period studied. the United States have been different if major farm The estimates of longrun and shortrun effects of farm commodity programs had been eliminated in 1953? To programs are extremely sensitive to changes in several help answer the question, an econometric model was assumptions that affect total supply and demand set up to simulate the behavior of selected economic elasticities in the model. Further, ordinary least squares variables during 1953-72.' regression analysis (OLS) was used to estimate the coef- Farm programs of the Federal Government have, in ficients of behavioral equations. Thus, the results should various ways, supported and stabilized farm prices and be considered preliminary and subject to revision if incomes since 1933, when the first agricultural adjust- alternative estimation techniques later reveal substantial ment act was approved. Since then, dramatic long-term differences for important coefficients. changes have occurred in (1) the resource structure of A central feature of the model—the disaggregation of agriculture into two sectors, crops and livestock—can be seen as both an advantage and a limitation. Use of two *Frederick J. Nelson is Agricultural Economist with the National Economic Analysis Division of the Economic sectors instead of only one does allow analysis of impor- Research Service. Willard W. Cochrane is professor of tant interrelationships between crops and livestock over Agricultural and Applied Economics at the University of time. But future research efforts should be aimed at a Minnesota. further extension to include specific commodities for ' A number of agricultural sector-simulation models devel- two reasons. First, persons and organizations that might oped in recent years can be used to quantify the total impact of be the most interested in the type of information availa- farm commodity programs. Some of these models were reviewed ble from the model would want answers for specific in this study (3, 8, 23, 24, 26, 30). The basic framework for this commodities. Second, commodity specific equations model resembles that in (30) and in (24). However, following Daly (2), a two-sector approach was used instead of the one- might provide more accurate quantitative results. For sector approach of Tyner (30) or the seven-sector method of example, measures of price variability for each com- Ray (24). modity are the most logical proxy measures of the

AGRICULTURAL ECONOMICS RESEARCH 52 VOL. 28, NO. 2, APRIL 1976 extent of risk and uncertainty. But they were not used • Given the above results, the model computes 1111) in the two-sector model.' various measures of income, price and income variability, and aggregate agricultural productivity. • Asset, investment, and debt levels, number of THE MODEL: ANALYTICAL FRAMEWORK, farms, and farmland prices are adjusted from the THEORY, AND SIMULATION PROCEDURE previous end-of-year levels, based on 1953 and earlier price and income experiences The analysis centers around a comparison of two • The above results are used to make similar calcula- simulated time series for each of several variables in tions for 1954 and later years given the complete 1953-72. One series shows estimates of the variables' time series for those explanatory variables not actual historical value with programs; the other, esti- determined within the model. mates in a free market without programs. The impact on The data used to measure the variables are based on a particular variable is the difference between its histori- published and unpublished calendar year information cal and free-market values, shown as a percentage from the Economic Research Service, and the Agricul- change in table 4 and figure 1 (see p. 59). tural Stabilization and Conservation Service in the U.S. As a measure of alternative impacts possible, several Department of Agriculture. However, only a few of simulation results were obtained, based on differing these variables are published in the exact form used here. assumptions about demand elasticities and resource To facilitate analysis, assets, inputs, production, and use adjustment responsiveness in a free market. This pro- statistics were measured in 1957-59 dollars; for price vided a test of the sensitivity of the model's results to indexes, 1957-59 equal to 100 was generally used. such changes. Detailed discussion is limited primarily to one simulation set. Farm Program Variables The farm programs covered include those involving Overview price supports, acreage diversions, land retirement, and foreign demand expansion. Programs involving domestic The simulation model consists of 59 equations (33 demand expansion, marketing orders and agreements, identities and 26 behavioral equations) and contains 51 import controls, and sugar are not explicitly included. exogenous variables.' A resource adjustment approach The programs included have affected agriculture in the to crop and livestock output and supply response was past two decades by: used in designing the model. The simulation procedure • Idling up to 16 percent of cropland (6 percent of •for each year is as follows (the calculation for 1953 is land in farms) through programs involving long- used as an example): and short-term acreage diversions to control out- • Current input levels are determined for the initial put year (1953) based on beginning-of-year asset • Diverting up to 16 percent of crop output from levels, current and recent price and income experi- the market into Government inventories or subsi- ences, and farm programs in use dized foreign consumption through price support • Crop productivity and production are determined and demand expansion activities endogenously, based on the level and relative • Providing farmers with direct Government pay- importance of selected inputs assumed to be pri- ments equal in value to as much as 29 percent of marily used for crop production net farm income (7 percent of gross income). • Crop and livestock supply and demand compo- Table 1 contains values of the exogenous farm program nents (including livestock production) and prices variables used. Table 2 shows the relative importance of are simultaneously determined once crop produc- some of these variables in the crop sector. The following tion is known and Government market diversions three sections explain more about use of these variables under the farm programs are specified and indicate the level for each program variable in the free-market simulation.'

2 Ray's disaggregation approach (24) is one alternative. Separate resource adjustment equations and production functions are included for livestock products, feed grains, wheat, An argument can be made in favor of making some or all soybeans, cotton, tobacco, and all other commodities. How- program variables endogenous. For example, CCC inventory ever, a procedure 4hat places less strain on the available data changes and acreage diverted by programs are complicated would be one that uses commodity acreage and yield equations functions of announced price supports (loan rates), diversion "controlled" by simulated aggregate resource and resource pro- requirements, and other supply and demand variables. Thus, ductivity adjustment estimates. See (22, p, 10; 34). exogenous price supports, instead of exogenous CCC inventory changes, could be used to represent the price support through 3 For a complete discussion of the theory, model, data, and simulation procedure, see (19). This information will also be acquisition and disposition activities of the CCC (as in (3)). available later in a planned USDA technical bulletin. A descrip- Further, one might want to specify only policy goals (such as tion of the variables and a list of the actual model equations are net income) as exogenous so that program operation rules would 1110 available from the senior author on request. need to be endogenous to determine program details each year in 53 Table 1.-Government farm program variables, 1950-72

Percent- Net Government Exports under Govern- Acres Percent- age of (CCC) inventory specified Govern- ment Direct of age of acres increases ment programs assisted Govern- cropland land in planted (1957-59 dollars) (1957-59 dollars) crop ment Year idled farms with exports farm by not hybrid Crops Livestock Crops Livestock (1957-59 program programs idled seed (CCCD) (CCLD) (GCX) (G LX ) dollars) payments (AD) (PCT) (PCTHB) (ASCX) (GP)

Millions Ratio Billion dollars

1950 0.0 1.0000 0.1900 -0.765 0.035 a 0.283 1951 0.0 1.0000 .1960 -.446 -.122 a a .286 1952 0.0 1.0000 .2010 .351 0.0 0.386 a 0.426 .275 1953 0.0 1.0000 .2040 2.164 .315 .369 0.063 .353 .213 1954 0.0 1.0000 .2060 1.028 .127 .531 .127 .319 .257

1955 0.0 1.000 .2130 1.289 -.203 .759 .214 .316 .229 1956 13.6 .9983 .2160 -.312 -.149 1.268 .231 .543 .554 1957 27.8 .9765 .2200 -.919 .051 1.219 .170 .933 1.016 1958 27.1 .9772 .2370 1.350 -.089 .978 .122 .737 1.089 1959 22.5 .9806 .2790 .282 -.031 1.030 .076 .775 .682

1960 28.7 .9753 .2910 .261 .049 1.351 .046 1.098 .702 1961 53.7 .9538 .2490 -.087 .113 1.308 .067 .950 1.493 1962 64.7 .9439 .2570 .191 .172 1.220 .089 .675 1.747 1963 56.1 .9514 .2750 -.016 -.103 1.227 .153 .755 1.696 1964 55.5 .9511 .2590 -.249 -.191 1.377 .176 .935 2.181

1965 57.4 .9500 .2600 -.532 -.031 1.183 .105 .780 2.463 1966 63.3 .9443 .2660 -2.008 -.037 1.214 .063 .923 3.277 1967 40.8 .9635 .2800 -1.192 .143 .920 .108 .783 3.079 1968 49.3 .9561 .2630 1.521 -.011 .870 .116 .528 3.462 1969 58.0 .9477 .2700 1.028 -.061 .711 .093 .550 3.794

1970 57.1 .9483 .2740 -.928 .010 .723 .070 .942 3.717 1971 37.2 .9663 .2970 -.213 -.007 .687 .096 .987 3.145 1972 62.1 .9433 .2740 -.862 -.008 .701 .044 1.137 3.961

allot available or not yet estimated.

Government market diversions. The Federal Government • CCCD is net stock change for crops owned by or supports farm commodity prices through operations of under loan with the CCC USDA's Commodity Credit Corporation (CCC). The • CCLD is net stock change for livestock products CCC helps farmers in three ways. It buys or sells com- owned by or under loan with the CCC modities on the open market, and extends loans to • GCX is crop exports under specified Govern- farmers who have the option of repaying the loan or ment programs delivering their commodity to the CCC in lieu of repay- • GLX is livestock exports under specified Govern- ment. Also, the CCC encourages domestic and foreign ment programs consumption by subsidizing food use or by giving com- • ASCX is crop exports assisted by the payment of modities away. Five exogenous variables represent this export subsidies by the CCC activity in the model: In the free-market simulation, these variables have a value of zero. Acreage diversions and Government payments. Farm the simulation. In the model, however, the procedure is to deter- program operations aimed at controlling supply-to mine the impact of program operations, not policies, with such operations defined in a special way. The total impact of past reduce the need for costly Government market diver- program operations is the main goal rather than the effect of sions-include offering farmers some combination of selected adjustments to specific annual policy variables or policy direct cash payments and price support through CCC goals. See (19, pp. 139-149). loan privileges in return for their idling of productive • 54 Table 2.-Farm program operations affecting crop output and marketings, 1950-72a • Acres diverted as Market percentage of diversions Total Crop- as Government Total land Total Total Land in Crop- percentage Year market acreage plus land in crop land of diversionsb diversions diversionsb farms production farms production

ercenr Billion dollars Million acres Billion dollars

17.0 0 0 d 1950 d 0 377 1,202 17.5 0 0 d 1951 d 0 381 1,204 18.4 0 0 7 1952 1.2 0 380 1,205 18.2 0 0 16 1953 2.9 0 380 1,206 17.9 0 0 11 1954 1.9 0 380 1,206

18.2 0 0 13 1955 2.4 0 378 1,202 18.3 1 4 8 1956 1.5 14 383 1,197 18.0 2 7 7 1957 1.2 28 386 1,191 19.9 2 7 16 1958 3.1 27 382 1,185 19.7 2 6 11 1959 2.1 23 381 1,183 20.8 2 7 13 1960 2.7 29 384 1,176 20.4 5 14 11 1961 2.2 54 394 1,168 20.7 6 16 10 1962 2.1 65 396 1,159 21.5 5 14 9 1963 2.0 56 393 1,152 20.7 5 14 10 1964 2.1 56 391 1,146 22.1 5 14 6 1965 1.4 57 393 1,140 21.6 6 16 1 1966 0.1 63 395 1,132 22.5 4 11 2 1967 0.5 41 381 1,124 23.2 4 13 13 1968 2.9 49 384 1,115 23.5 5 15 10 1969 2.3 58 391 1,108 22.6 5 15 3 • 1970 0.7 57 389 1,103 25.1 3 10 6 1971 1.5 37 377 1,097 25.3 6 16 4 1972 1.0 62 398 1,093

aThe information does not represent precise estimates of "excess capacity" in U.S. agriculture, but rather a summary of some relevant magnitudes. These do, of course, have implications for excess capacity analysis. bGovernment market diversions include the sum of net change in Government crop inventories (CCCD), Government crop exports (GCX), and assisted commercial crop exports (ASCX). clncludes acres of cropland harvested, crop failure acreage, cultivated summer fallow acres, plus acreage diverted by farm programs (AD). dNot available or not yet estimated.

cropland. The acreage idled under annual diversion and grams began that year and they affected the relative long-term land retirement programs (AD) is included as importance of corn and sorghum acreage. Therefore, in an explanatory variable in the equation for the use of the free-market simulation, PCTHB was assumed to cropland. The associated Government payments (GP) are increase a little faster from 1956 to 1959 than in actual included as part of gross and net farm income. In the history. The record level of PCTHB for 1971 (0.297) free-market simulation, both of these variables have a was assumed to have been achieved throughout 1961-72, value of zero. The percentage of total cropland not idled after the high level achieved in 1960 (0.291.)' (PCT) is used in the analysis; its free-market value is, of course, 1.0 (100 percent). Cropland planted with hybrid seed. The increased use Following the theoretical ideas of Griliches (7), one could of high-yielding corn and sorghum grain seed has been an argue that the percentage of cropland planted with hybrid seed important technological advance on American farms. should be endogenous because the corn price level affects the The percentage of total cropland planted with hybrid profitability of adopting more expensive, higher yielding seed. An adequate consideration of this question will have to wait seed (PCTHB) is used as an exogenous explanatory until commodity specific extensions are made. The percentage variable in the fertilizer and crop productivity behavioral for all cropland depends on the relative importance and geo- equations. It was assumed that the upward trend in graphic location of corn and sorghum acreage as well as on faCTHB was retarded in 1956 because acreage-idling pro- prices received for corn and sorghum. 55 Special Features • Minimizing the probability that farm losses will Current input and asset adjustment. Behavioral equa- lead to financial disaster by reducing the total et tions representing the demand for assets were specified amount of investment in the farm business which assuming asset adjustments occur in response to changes reduces the potential size of both profits and in (1) longrun profit expectations and (2) the extent of losses risk and uncertainty. Separate equations were included • Increasing the firm's ability to survive loss experi- for the quantity of land and buildings, machinery and ences by increasing the share of total farm business equipment, and livestock number inventories. The stock investment held as financial reserves and operating of an asset is determined by its level in the previous year, with smaller amounts of borrowed capital. with adjustments for depreciation and for investments. (Elements of the first two adjustments may be involved A partial resource adjustment assumption was used in when farmers choose to participate in specific voluntary specifying demand equations for assets based on the Ner- price support-acreage diversion programs.) Because of lovian distributed lag procedure. Longrun demand was the desire for financial reserves, an important interrela- explained by including as variables current and recent tionship probably exists between annual investments, factor-factor price ratios, relative rates of return to farm savings, family consumption, and risk and uncertainty. A real estate, and risk and uncertainty proxy indexes. realistic appraisal of the economic consequences of Current input expenditures depend on current and eliminating price stabilizing programs must consider this recent factor-product price ratios, asset levels, other factor of farmers' risk aversion.' input levels, and risk and uncertainty proxy indexes. The Proxy indexes of the extent of risk and uncertainty model contains behavioral demand equations for the were computed in the model as 5-year averages of the following current inputs to agriculture: repair and absolute annual percentage change in prices and in operation of machinery, repair and operation of incomes. These indexes were included as explanatory buildings, acres of cropland used for crops, fertilizer and variables in the behavioral equations for assets and lime, crop labor, livestock labor, hired labor, and miscel- inputs. Proxy indexes were computed for the following laneous inputs. The use of "other" input and asset levels variables: (1) aggregate crop price index, (2) aggregate as explanatory variables in current input demand func- agricultural price index, (3) net income available for tions is consistent with traditional profit-maximizing investment (net income plus depreciation allowances), theory, because the marginal product of one factor and (4) the livestock-crop price ratio. Direct Govern- depends on the quantity used of other factors. In the ment program payments to farmers (GP) were also used short run, current inputs adjust toward longrun levels as to explain resource adjustments; GP was assumed to 41) asset adjustments occur. Use of other current inputs as represent a relatively certain source of net income for explanatory variables in the input demand functions the coming year, once the annual program details had resulted in a set of simultaneous equations. been announced by USDA. Price and income expectations, and risk and uncer- Behavioral equations for the following variables con- tainty. Price and income expectations were represented tain one of the several risk and uncertainty proxy varia- by including current or lagged values of prices and bles: repair and operation of machinery, fertilizer and income in input and asset adjustment equations. Simple lime, acres of cropland, repair and operation of build- averages of up to 5 years were sometimes used if more ings, miscellaneous inputs, buildings, land in farms, live- than one observed value was assumed relevant. stock number inventory, and farmland prices. Demand A major assumption was that an increase in com- equations for machinery, labor, and onfarm crop inven- modity price variability specifically, and the elimination tories contain no risk proxies. of farm programs generally, would increase the risk of Crop input and productivity. Crop output is the investing in agriculture. Therefore the level of invest- product of three variables: ment and current input expenditures for any given level • Sum of four inputs (measured in 1957-59 dollar of average price and income expectations would be values) used primarily for crop production— reduced. The idea behind the assumption is that farmers fertilizer and lime, machinery inputs, acres of will adjust to situations involving varying degrees of cropland for crops, and man-hours of crop labor price and income uncertainty by sacrificing some poten- • Percentage of cropland harvested (exogenous) tial profits to reduce the probability of financial disaster. • Output per unit of crop input Such adjustments depend on a farmer's psychological In specifying an output per unit of crop input equation, makeup and capital position, and they can take several forms:

• Adjusting the planned product mix to favor 'This explanation follows Heady's (11, pp. 439-583). Sup- products with relatively low price and income port also appears in (6, 9, 15, and 16). And see the recent quan- variability titative analysis of farmer investment and consumption behavior • reported in (5), also an empirical test of the hypothesis that Diversifying in a way that reduces net farm income farmers' cropping patterns and total outputs are influenced by variability a consideration of risk as well as expected income in (18). e

56 crop productivity increases specifically, and farm tech- reflected in a reduced level of nonfarm purchased inputs nological advances generally, were assumed to have relative to land and labor. As a result, aggregate crop geoccurred along with, or partly because of, the greater resource productivity went down and crop and livestock use of nonfarm produced inputs relative to the tradi- prices increased over time. Further, as prices rose in the tional inputs of land and labor. model, additional cropland and other crop inputs were Farm technological change can be seen as the longrun pulled into the system. But average crop input produc- result of specialization of labor and the associated highly tivity was further decreased, which tended to dampen successful innovative effort and research investment by the supply response and retard the expected downward persons in both the public and private sectors. The farm pressure on prices. This illustrates the advantage of input and public sectors of the economy have become endogenously simulating productivity in preference to specialized producers of a continuous stream of new using a simple extension of past trends. improved products and technologies that are used by Supply, demand, and prices. Total supplies of crops farmers. Farmers, in turn, have become specialists in and livestock were set as identically equal to current organizing and using these products so that inputs of production, plus beginning-of-year private stocks, and land and human capital have become more productive. imports (for livestock, minus exports). The associated These changes have resulted mainly in response to demand components include feed, seed, domestic human economic incentives and they involve dynamic adjust- consumption, commercial exports, exogenous exports ments in the demand and supply of technology. Farmers assisted by export subsidies or other specified Govern- have demanded improved inputs and techniques to maxi- ment programs, exogenous CCC net inventory changes, mize profits. And suppliers have developed the new and end-of-year private stocks. Measures of "open- products and techniques desired. Farm technological market," or "commercial," supply were defined as total change depends on resource substitutions and capital supply minus Government market diversions (CCC net outlays by farmers in response to: inventory changes plus Government-aided exports). • Changes in factor and product price relationships Given the level of crop production, the supply and • Cost and availability of new inputs and techniques demand equations are used to simultaneously determine • Expected benefit from adoption of new inputs and livestock production, livestock and crop prices, and the methods endogenous components of demand. Each such com- • Farmers' liquid and capital assets position ponent is, directly or indirectly, a function of beginning- • Extent and importance to farmers of risk and of-year private stocks, population, disposable personal uncertainty' income per capita, a nonfood price index, the various The output per unit of crop input index was esti- exogenous Government market diversion variables, • mated as a linear function of several 9fiables: exogenous crop exports and crop imports, crop produc- • Percentage of cropland planted with hybrid seed tion, and a time trend. • Ratio of nonfarm produced fertilizer and Alternative simulation sets, or runs, discussed below, machinery inputs to crop labor and cropland were based on the use of alternative demand equations inputs for domestic human consumption (because these could • Crop inputs subtotal not be successfully estimated by usual regression analy- • Squared interaction term between the first two sis) and the use, in one simulation, of a synthesized items in this list. equation for the foreign demand for crops.8 (Input and output measures used are value aggregates Aggregate prices, incomes, and other equations. based on 1957-59 average prices.) The hybrid percentage Detailed results from preceding components of the was assumed to increase productivity because of the model are used to compute an index of agricultural tremendous yield-increasing effect of shifts to hybrid prices, various measures of income (including gross and corn and sorghum seed. Productivity was assumed to net farm income and the rate of return in agriculture decline as total inputs increased, because, for example, relative to the market interest rate), and several measures greater land use would likely extend to less productive of price and income variability assumed to reflect the cropland. The ratio of nonfarm inputs to land plus labor extent of risk and uncertainty. The quantity of hired was assumed to increase productivity. In the analysis of farm labor and the hired farm wage rate are determined farm program impacts, this crop productivity equation simultaneously. From these results, farm production significantly helped to explain longrun price trends and expenses for labor and a residually computed family cycles. Because of the method used to specify the crop labor input are derived. Farm prices and the nonfarm productivity equation, financial losses and business disasters simulated in the free market were ultimately One set of domestic demand equations is based on the elasticity matrix of (4). Another set is derived using simple 'These ideas are based on concepts in (1, 10, 27, and 6). analysis of the relationship between income-deflated price and Shortrun and longrun foreign The quantitative procedure used was influenced by the work consumption, used in (33). demand elasticities for crops are based on (28). _0 in (17, 21, and 32).

209-040 0 - 76 - 3 wage rate are two of the explanatory variables deter- mining the wage rate for hired labor. Farm land values Table 3.-Simulation alternatives and the number of land transfers per 1,000 farms are determined simultaneously. Farm prices, aggregate agri- Number fora Demand assumption cultural productivity, and nonfarm price levels are three Historical Free-market of the variables used to explain land values.' simulation simulation Output per unit of input for the total agricultural sector is derived from estimates of crop and livestock Least inelastic demand assumption b 13 14 production and from the inputs previously estimated. Moderately inelastic Other equations included in the model compute demand assumption c 18 19 (1) the number of farms, based on an estimate of average Most inelastic demand farm size, (2) gross farm capital expenditures, (3) farm assumption d 9 10 debt, and (4) total quantity and current value of assets. aThese numbers identify the alternative simulations in the text, table, and charts of this article. bDomestic demand Simulation Procedures and Alternatives equations were based on domestic demand for human consumption elasticities shown in (4, pp. 64-66 and Results for three alternative simulation sets are dis- 46-51). Own elasticities for domestic consumption of cussed below.' ° Each set includes a simulation of a crops and livestock are -0.274 and -0.259 respectively. free market situation and the actual historical situation. Commercial crop exports were made endogenous by using These alternatives were developed because of the diffi- foreign demand elasticities based on those reported in (28). The foreign demand elasticities are -1.0 in the short culty of estimating theoretically correct demand equa- run and -6.0 in the long run. cSame domestic demand tions for domestic human consumption and crop ex- parameters discussed in previous footnote, but commer- ports by usual procedures. The three sets appear in table cial crop exports were made exogenous and equal actual 3, and its footnotes describe the procedure and sources historical levels. dCrop exports were considered exoge- briefly.' nous, as in footnote three, but domestic demand functions were derived by graphic analysis of the relationship between income deflated price and per capita consumption during the period. (See (33, pp. 11-18), for example). Here, own elasticities are -0.11 or -0.15 for livestock and 'Equation specifications were influenced by (31) for hired -0.07 or -0.13 for crops. labor and (14) for land prices. 1°The computer simulation procedure uses the Gauss-Seidel algorithm to obtain a solution of this nonlinear system by an EFFECTS OF ELIMINATING FARM iterative technique (13). Bob Hoffman and Hyman Weingarten, COMMODITY PROGRAMS IN 1953 ERS, made programming revisions needed to facilitate use of the Gauss-Seidel procedure. What would have happened in American agriculture Six additional simulation alternatives appear in (19, p. 232, table 19). These are based on arbitrary revisions in the resource had farm programs been eliminated in 1953? Some adjustment equations made to allow for possible additional possible answers to this question are provided by the effects of increased risk and uncertainty in a free market. results in table 4 and figures 1-8. One measure of the

Table 4.-Effects on selected variables of eliminating farm programs in 1953, five-year averages, 1953-72a

Percentage change from historical value Item 1953-57 1958-62 1963-67 1968-72 Crop supply to open market (CSPLY)b 8.4 2.6 -4.3 -9.5 Livestock supply to open market (LSPLY)b 3.8 4.8 3.4 -3.9 Price index for crops (PC) -28.2 -22.6 -8.1 31.7 Price index for livestock (PL) -19.5 -25.8 -18.5 25.2 Price index for agriculture (PA) -23.2 -24.4 -14.9 27.7 Total net income (TN I) -42.0 -37.7 -19.7 40.3 Total agricultural productivity index (TLB) 1.5 3.7 2.4 -5.1 Price index for land and buildings (PLD) -4.6 -12.4 -16.8 -16.5 Gross farm capital expenditures (GCE) -20.9 -54.3 -47.3 -12.7 Total production assets at end of year (ASSET) -1.7 -7.0 -10.0 -10.0 Agricultural price variability index (SPA) 52.7 7.2 36.1 150.0

a Based on results of simulations 18 and 19, which use demand parameters derived from demand matrix in (4). Exports are assumed to be exogenous. bSupply includes production minus GoveMment market diversions plus beginning-year private stocks plus net private imports for livestock and gross imports for crops.

58 FERTILIZER AND LIME PERCENTAGE CHANGE IN INPUTS: AGRICULTURAL PRICE INDEX: HISTORICAL TO FREE-MARKET LEVEL ACTUAL AND SIMULATED VALUES

PERC ENT $ B IL.

50 3 25 _ . 0 2 . / ..••• . ...• / .... •ft. • ••• ..... _ ...... • „,..• -25 _ ..• . _ . — ../ ' 1 -50 _ -75 1 1 11 0 'III Jill 1 1 I 1 1111 FIGURE 1 FIGURE 2 1952 '57 '62 '67 '72 MACHINERY INPUTS: COMPARED WITH INELASTIC ACTUAL AND SIMULATED VALUES SIMULATION SIMULATION DEMAND $ IL. — 10 9 Most — - 19 18 Moderately 8 --- 14 13 Least Note: See Table 3 for explanation of alternative simulations. 7

•■ • 6 ..... ;0.* ...... / impact of farm programs on a variable is the difference - .... / • between the simulated historical level and the simu- _ •.... till I 1 I I ""'r -1-1° • 1 till lated free-market level. Such differences are shown in 5 • FIGURE 3 figure 1 and table 4 as percentage changes from the historical to the free-market levels. CROPLAND INPUT INDEX: Alternative Impacts on Prices ACTUAL AND SIMULATED VALUES The impacts of eliminating farm programs, on agri- $ BIL cultural prices, for the three alternative simulation sets discussed in table 3, are shown in figure 1. The patterns - \ . of percentage impacts on prices for each demand alterna- 1.20 . . .... .•••■/•• tive resemble one another to some extent. Each is - % / • ■ initially negative and each grows over time until the ' **>," \ largest negative impact occurs in 1957. Afterwards, the 1.15 magnitude reduces gradually as the free-market price _ I\ level becomes equal to and greater than the historical _ i \ A t I, i level by 1967. The largest positive impact occurs in 1.10 1969-71. However, the degree of impact differs importantly among the alternatives in most years, a behavior / 1 that highlights the important interrelationship between _ ,, 1.05 . , , . , i , i , i , t t , , i the assumed elasticity of demand and the estimated FIGURE 4 impacts of the farm programs. Under all three demand alternatives, it is estimated 1952 '57 '62 '67 '72 that prices in the free market would have been lower than in actuality during 1953-65. By 1957, the reduc- — ACTUAL tion would have been 20 percent for the least inelastic -- HISTORICAL" FREE-MARKET* demand assumption, 33 percent for the moderately inelastic demand assumption, and 54 percent for the "Historical simulation 18 : free-market simulation 19.

59 TOTAL AGRICULTURE PRICE INDEX FOR PRODUCTIVITY INDEX: AGRICULTURAL PRODUCTS: ACTUAL AND SIMULATED VALUES ACTUAL AND SIMULATED VALUES

% OF 1141.11 %O tau 1 -uu Hi/1]. .../ ..., 150

100 •rr • 1%.•••■••••••"-'>41.- -.....e-.' ./ . .r . 100 \ "•••••• ...... • / .."

50 80

0 III I 1 I I I ill' ti ll FIGURE 6 60 I I I 1111 1111 1111 FIGURE 5 RELATIVE RATE OF RETURN TO FARM REAL ESTATE: TOTAL NET INCOME ACTUAL AND SIMULATED VALUES INCLUDING NET RENT: RATI ti ACTUAL AND SIMULATED VALUES - - e -N' . " • Sal 1.6 / / N. -.•"- .- .." st ....* : N... I Ni... - .0.• N • 0.8 1.• 20 - \ : i " ....N. ...•/".... / ...... ,_. , / •/ ....'". / 0 ‘...... \ ....\• .....d. .--- ....- 10 _ \ ... • -0- •• `...... -0.8 M i t 1 i i 1 I t i i 1 t 1 t 1 FIGURE 8 1 i 11 1111 1111 1111 0 FIGURE 7 1952 '57 '62 '67 '72

1952 '57 '62 '67 '72 --- ACTUAL -- HISTORICAL* — ACTUAL FREE-MARKET* -•- FREE-MARKET* -- HISTORICAL* *Historical simulation 18 : free-market simulation 19. "Historical simulation 18 : free-market simulation 19.

most inelastic demand assumption. In all three cases, keep them lower than otherwise during 1968-72. prices would have begun to recover after 1957, but This latter result differs importantly from those in would not have returned to their actual historical levels other historical free-market studies. For example, Ray until around 1967, 10 years after the 1957 low and 14 and Heady report that low free-market prices would years after the programs had been eliminated. Prices have depressed income and increased supplies through- would have continued to increase, relative to the his- out their period of analysis-1932-67 (25, p. 40). In torical situation, until they peaked during 1969-71. Tyner and Tweeten's study, prices are lower in the free- Eliminating farm programs in 1953 would have raised market simulation than in the historical simulation for 1972 farm prices 6 percent under the least inelastic all periods reported-1930-40, 1941-50, and 1951-60 demand assumption, 35 percent under the moderately (30, p. 78). In both studies, the supply response in inelastic demand assumption, and 68 percent under the agriculture is never enough for free-market farm prices most inelastic demand alternative. Thus, farm programs to recover fully. One explanation is that the rate of kept farm prices higher than they otherwise would have technological advance was exogenous in the previous ilk been during 1953-65, but the cumulative effect was to models while in this model, such change is endogenous.

60 Results For Moderately Inelastic inputs would have been used by 1972, in the free market. Demand Alternative But crop productivity would have dropped 19 percent Effects of eliminating farm programs in 1953 are also below the actual historical level, cutting crop production presented in table 4 and figures 2-8. These results are 13 percent. based on a comparison of historical simulation 18 and Farm income. Total net farm income, in the free free-market simulation 19.' 2 This set of results is not market, would have averaged 42 percent below histori- necessarily the "best," or "most correct." It was selected cal levels in 1953-57. Such income would have been 20 primarily because the results represent a kind of mid- percent below the actual level in 1953. By 1957, income range between the alternatives, as indicated in figure 1. would have dropped $8 billion, to equal 55 percent of Presenting only one set of results facilitates understand- actual income that year. Further, though net farm ing the dramatic and interrelated effects that would have income would have remained more than $3 billion lower occurred in the absence of the programs. through 1966, it would have finally risen to a level Supplies and prices. Changes in the aggregate farm nearly $10 billion higher than historical levels in 1971 price level for the free-market situation, compared to and 1972. Such income would have climbed 58 percent actual history, resulted primarily from changes in crop above the historical level in 1971, to average 40 percent supply and price. As one might reasonably expect, crop higher during 1968-72 (fig. 7). price adjustments also determined eventual livestock Figure 8 shows the impact of eliminating farm pro- price adjustments. Over time, livestock producers adjust grams on the rate of return to farm real estate (relative their inventory and production levels in response to to market interest rates). Residual returns to real estate changes in the livestock-crop price ratio. Crop price in a free market would have been negative in 1954-62, changes were determined mainly by changes in open making estimated losses comparable to those in the de- market crop supplies tempered by simultaneous adjust- pression years, 1930-33. As with price and net income, ments in feed use and private end-of-year inventory the rate of return in a free market would have been levels. higher than its historical level after 1967. However, the Actual crop prices were significantly affected by large highest free-market rate of return ratio (RATO=2.0 in Government market diversions equal to over 10 percent 1969) would not have been as high as that for the war- of actual production in 1953-55. With price-supporting influenced period of 1942-48, when the ratio varied activities eliminated in 1953, crop prices would have from 2.1 to 3.8. fallen sharply as stocks increased in the short run. In a Assets, investments, and land prices. Assets, value of free-market situation, private crop stocks would have capital expenditures, and land prices would all have been been 17 percent higher than the historical level in 1955, lower in a free market than historically for 1953-72 and crop prices, 36 percent lower. Open market crop (table 4). Low prices and incomes and increased risk and supplies would have continued to exceed historical uncertainty would have immediately and subsequently supply levels throughout 1955-64, because crop produc- affected the amount of assets farmers would have been tion decreases would not have been large enough to off- willing and able to buy. Gross farm capital expendi- set the effect of elimination of Government market tures would have declined dramatically. Reaching a level diversions. Actual diversions, substantial in this period, 59 percent below actual historical levels by 1960, they ranged from 7 to 16 percent of actual crop production, would not have returned to a point near actual levels though 4-16 percent of the cropland was idled by until 1971 and 1972. Total productive assets in a free existing programs. After 1964, however, crop produc- market would have averaged 10 percent below actual tion decreases in a free market would have become larger historical levels during 1963-72, and farm land prices than actual Government market diversions under the would have averaged 17 percent below actual values. program. Thus, free-market crop supplies would have Agricultural productivity. The agricultural produc- fallen below historical levels in 1965; and, by 1972, they tivity index would have been somewhat higher in a free would have been down 11 percent. Crop prices would market than it actually was from 1955 to 1968, reaching have been 36 percent higher in 1972 than they actually a high of 7 percent more in 1958. However, the longer were in that year. term effect of eliminating farm programs would have The relative decrease in crop production after 1964 been to reduce the productivity index to a level 11 would have dramatically affected farm prices throughout percent below the historical level by 1972. In 1961, the 1964-72 (fig. 6). As a result, 8 percent more crop related index would have been 101 (1967 = 100), never to exceed 102 in subsequent years of the free-market

' ' Historical simulation 18 can also be compared with the simulation (fig. 5). actual variable values plotted in figures 2-8. However, some Crop productivity in a free market would have fallen equations have been adjusted to reproduce history more accu- below actual historical levels for all years after 1958, and rately than otherwise through use of regression error ratios. would have been down 19 percent by 1972. Most of Such adjustment was considered desirable because the model is this 19-percent decrease would have been attributable to nonlinear. Thus, important disturbances in the equations could affect accuracy of the estimated program impacts. the decline in use of nonfarm inputs (such as fertilizer

61 and machinery) relative to cropland (figs. 2-4). The ratio market while shares for all other assets would have of machinery and fertilizer to land and labor would have declined. Crop labor requirements would have risen been 52 percent lower in the free market situation.' from 7 to 15 percent of total current inputs. Cropland Also, the increased use of lower quality land would have • would have changed from 3 to 4 percent; livestock labor, reduced crop productivity; but an increase in the relative from 4 to 5 percent. Other input shares would have use of hybrid seed would have raised productivity. declined. Decreased machinery inputs and increased use of crop- Agricultural employment would have risen, with land would have substantially raised labor inputs for labor requirements 73 percent higher in 1972 than with 1957-72 in a free market. farm programs. Most of the increased labor would have Agricultural price variability. Absolute annual per- come from farm operators or their families. Family labor centage changes in the agricultural price index would would have gone up 120 percent but hired labor inputs have averaged substantially above historical levels in a would have gained only 19 percent. free-market situation. For the initial 5-year period, 1953-58, this index of variability would have averaged 53 percent higher. It would have continued above his- ASSESSMENT torical levels for all but 2 years. By 1968-72, the index would have averaged 150 percent higher. The following summarizes results from simulations Organization and structure. Several organizational using demand relationships implying an aggregate and structural changes in agriculture would have domestic demand elasticity of around -.25 and assuming occurred had farm programs been eliminated in 1953. commercial crop exports are fixed at their actual histori- Number of farms would have risen while the average size cal levels in the free-market case (simulations 18 and 19). dropped. Land in farms relative to other assets would These results suggest that at least seven different impacts have increased, and cropland and labor would have been on the agricultural economy would have occurred had substituted for machinery and fertilizer inputs. farm commodity programs of the Federal Government In the free market, the number of farms would have been eliminated in 1953: declined, but not as fast as it actually did. In historical • Farm prices would have dropped for several con- simulation 18, number of farms declined at the average secutive years until they averaged 33 percent be- annual rate of 3.0 percent per year to a 1972 level of low actual levels by 1957 2.7 million. In free-market simulation 19, the number of • Aggregate farm prices would have been stable but farms declined at the rate of 1.9 percent per year to 3.3 low until after 1964, when they would have risen million in 1972. (The simulated number of farms was to a level averaging 35 percent above the actual 4.7 million for 1953.) In 1972, there would have been figure in 1972 • 24 percent more farms than in actual history because the • Net farm income would have fallen 55 percent average size would have been 19 percent lower while below the actual level by 1957 but it would have total land in farms remained essentially unchanged. reached 58 percent above the actual level in 1971 (Elimination of farm programs did affect land in farms • Residual returns to owners of farm real estate prior to 1972.) would have been negative in 1954-62 Average farm size in 1972 would have been much • Quantity of assets, value of capital expenditures, lower in a free market because agriculture would have and farmland prices all would have been lower been less mechanized, with more labor used per acre. A than &xial levels throughout 1953-72, as a free market from 1953 on would have slowed the rate at result of farmers' response to the initial and sub- which machinery and fertilizer and other nonfarm pro- sequently lower price and income experiences, duced inputs were substituted for land and labor. Thus, lower expectations, and increased risk and uncer- farmers would have had less inducement to reorganize tainty operations into larger sized units. In the historical simu- • Land and labor inputs would have increased rela- lation, the average size of farm increased at the average tive to other inputs, and the rate of decline in rate of 2.5 percent per year from 1953 to 1972. In the agricultural employment and number of farms free market, this figure would have been 1.4 percent. during 1953-72 would have been reduced The share of total assets made up by land would have • Crop resource productivity would have dropped increased from 55 percent to 60 percent with a free under historical levels in all years after 1958, to be down 17 percent in 1972 • Agricultural productivity (crops and livestock A net decrease in crop productivity in this free-market combined) would have been 11 percent under simulation results mostly from the effect of reduced machinery actual levels in 1972. relative to cropland and labor. The effect of less use of machin- Thus, farm programs had substantial and important ery offsets a technically inappropriate positive effect of reduced fertilizer. The fertilizer sign comes from a negative partial deriva- effects on the developments in the agricultural sector tive of productivity with respect to fertilizer of -0.1 obtained for during the period studied. In particular, the programs the crop productivity equation. apparently worked to promote both long- and short- ilk.

62 term price and income stability. Apparently, the poten- result in a free market. Without farm commodity pro- tial exists for continuous long-term food and fiber price grams, consumers would have enjoyed low farm product cycling because of the nature of agricultural supply prices through 1964. Farmers, at the same time, would responses in a free-market situation. This cycling would have suffered their worst financial crisis since the Depres- occur, as the domestic and world economies grow, be- sion. But these low prices would have been replaced by cause domestic agricultural supply cannot grow at high farm prices, following a long period of rapid farm exactly the same rate as demand. The growth rate for price increases after 1964. At the same time, farm supply is affected by complex interrelationships that incomes would have been improved greatly. exist between (1) adjustments in agricultural assets and inputs, in response to price and income experiences, and (2) adjustments in crop productivity and livestock production. During 1953-72, farm commodity programs REFERENCES were operated in a way that mitigated aggregate farm price and income cycling over extended periods. (1) Cochrane, Willard. Farm Prices Myth and Reality. This study suggests that farm programs supported Univ. Minnesota Press, Minneapolis, Minn., 1958. farm prices and incomes at levels substantially higher (2) Daly, Rex F. "Agriculture: Projected Demand, than they would have been otherwise during 1953-65. Output and Resource Structure." Implications of Feed and other crop prices were supported by programs Changes (Structural and Market) on Farm Manage- that idled productive land and diverted marketable ment and Marketing Research. CAED Rpt. 29, Ctr. supplies into Government storage or that subsidized for Agr. and Econ. Adjustment, Iowa St. Univ. domestic and foreign use. This resulted in reduced Science and Technol., Ames, Iowa, 1967, livestock production and consumption, and higher live- pp. 82-119. stock prices. Farmers responded to these developments (3) Egbert, Alvin C. "An Aggregate Model of Agricul- by mechanizing, fertilizing, increasing farm size on the ture-Empirical Estimates and Some Policy Implica- average, and generally adopting technologies that re- tions."Amer. J. Agr. Econ. 51: 71-86, February duced costs, boosted resource productivity, and 1969. expanded productive capacity. Elimination of farm pro- (4) George, P. S. and G. A. King. Consumer Demand grams in 1953 would have slowed the rate at which for Food Commodities in the United States with these advancements took place, or reversed the trend Projections for 1980. Giannini Foundation Mono- temporarily. The result: in recent years (1968-72), farm graph 26, Univ. Calif., Davis, Calif., 1971. price levels would have been higher in a free market than (5) Girao, J. A., Tomek, W. G., and T. D. Mount. • in actuality. "Effect of Income Instability on Farmers' Con- Farm prices in the free-market simulation eventually sumption and Investment Behavior: An Econo- recovered, and finally exceeded actual historical levels, metric Analysis." Search, -Vol. 3, No. 1, Agr. because elimination of farm programs in 1953 put agri- Econ. 5, Cornell Univ., Ithaca, N.Y., 1973. culture through the "longrun wringer."' With free- (6) Gray, Roger W., Vernon L. Sorenson, and market prices 10 to 30 percent below actual levels Willard W. Cochrane. The Impact of Government throughout 1953-66, and a negative rate of return to real Programs on the Potato Industry. Tech. Bul. 211, Minn. Agr. Expt. Sta., Minneapolis, Minn. 1954. estate for a number of years, gross capital expenditures Griliches, Zvi. "Hybrid Corn: An Exploration in and current input expenditures were greatly reduced, (7) the Economics of Technological Change." Econo- and agricultural productivity and output growth retarded. metrica. Vol. 25, No. 4, October 1957, pp. 501-522. The eventual result in the free-market simulation was (8) Guebert, Steven R. "Farm Policy Analysis Within that farm prices increased dramatically as aggregate a Farm-Sector Model." Unpublished Ph. D. thesis, demand grew faster than aggregate supply. Farm com- Univ. Wisc., Madison, Wisc., 1973. modity programs held farm prices and incomes higher Hathaway, Dale E. Effect of Price Support Pro- than would have been true otherwise for 1953-65, which (9) gram on the Dry Bean Industry in Michigan. Tech. apparently provided the economic incentives to growth Bul. 250, Mich. St. Univ. Agr. Expt. Sta., East in output sufficient to hold farm prices lower than they Lansing, Mich., 1955. otherwise would have been for 1968-72. (10) Hayami, Yujiro and Vernon Ruttan. Agricultural These results suggest that the national agricultural Development: An International Perspective. John plant can and does respond to changes in economic Hopkins Press, Baltimire, Md., 1971. incentives, given sufficient time. But because substantial (11) Heady, Earl 0. Economics of Agricultural Produc- time is required to change agricultural capacity, long tion and Resource Use. Prentice-Hall, Inc., Engle- periods of substantial disequilibrium and disruption can wood Cliffs, N.J., 1952. (12) Heady, Earl 0., and Luther G. Tweeten. Resource ''Cochrane discussed how the "longrun wringer" could Demand and Structure of the Agricultural Indus- .40 "correct" the surplus condition in agriculture in (1, pp. 134-136). try. Iowa St. Univ. Press, Ames, Iowa, 1963.

63 (13) Helen, Dale, Jim Matthews, and Abner Womack. culture."Agr. Econ. Res. Vol. 24, No. 3, "A Methods Note on the Gauss-Seidel Algorithm July 1972. for Solving Econometric Models."Agr. Econ. Res. (24) Ray, Daryll E. and Earl 0. Heady. "Government • Vol. 25, No. 3, July 1973, pp. 71-75. Farm Programs and Commodity Interaction: A (14) Herdt, Robert W. and Willard W. Cochrane. "Farm Simulation Analysis."Amer. J. Agr. Econ. Vol. 54. Land Prices and Farm Technological Advance." No. 4, November 1972, pp. 578,590. J. Farm Econ. Vol. 48, No. 2, May 1966, pp. 243- (25) Simulated Effects of 263. Alternative Policy and Economic Environments on (15) Johnson, D. Gale. Forward Prices for Agriculture. U.S. Agriculture. Ctr. for Agr. and Rural Devlpt., Univ. Chicago Press, Chicago, Ill., 1947. Iowa St. Univ., Card Rpt. 46T, Ames, Iowa, 1974. (16) Johnson, Glenn L. Burley Tobacco Control Pro- (26) Rosine, John and Peter Helmberger. "A Neoclassi- grams. Bul. 580, Ky. Agr. Expt. Sta., Lexington, cal Analysis of the U.S. Farm Sector, 1948-70." Ky., 1952. Amer. J. Agr. Econ. Vol. 56, No. 4, November (17) Lambert, L. Don. "The Role of Major Technolo- 1974, pp. 717-729. gies in Shifting Agricultural Productivity." Unpub- (27) Ruttan, Vernon W. "Agricultural Policy in an lished paper presented at joint meeting of Amer. Affluent Society." J. Farm Econ. Vol. 48, No. 5, Agr. Econ. Assoc., Canadian Agr. Econ. Soc. and December 1966, pp. 1100-1120. Western Agr. Econ. Assoc., Edmonton, Alberta, (28) Tweeten, Luther G. "The Demand for United Aug. 8-11, 1973. States Farm Output." Stanford Food Res. Inst. (18) Lin, William, G. W. Dean, and C. V. Moore. "An Studies, Vol. 7, No. 3, 1967, pp. 347-369. Empirical Test of Utility vs. Profit Maximization (29) Tyner, Fred H. and Luther G. Tweeten. "Excess in Agricultural Production."Amer. J. Agr. Econ. Capacity in U.S. Agriculture." Agr. Econ. Res., Vol. 56, No. 3, August 1974, pp. 497-508. Vol. 16, No. 1, January 1964, pp. 23-31. (19) Nelson, Frederick J. "An Economic Analysis of (30) "Simulation As a Method the Impact of Past Farm Programs on Livestock of Appraising Farm Programs." Amer. J. Agr. and Crop Prices, Production, and Resource Adjust- Econ. Vol. 50, No. 1, February 1968. ments." Unpublished Ph.D. thesis, Univ. Minn., (31) Tyrchniewicz, Edward W., and Edward G. Schuh. Minneapolis, Minn., 1975. "Econometric Analysis of the Agricultural Labor (20) Penson, John B., Jr., David A. Lins, and C. B. Market." Amer. J. Agr. Econ. Vol. 51, No. 4, Baker. "The Aggregative Income and Wealth November 1969, pp. 770-787. Simulator for the U.S. Farm Sector-Properties (32) Ulveling, Edwin F., and Lehman B. Fletcher. "A • of the Model." Unpublished ms., National Econ. Cobb-Douglas Production Function with Variable Analysis Div., Econ. Res. Serv., U.S. Dept. Agr., Returns to Scale." Amer. J. Agr. Econ. Vol. 52, Washington, D.C., November 1974. No. 2, May 1970, pp. 322-326. (21) Quance, C. Leroy. "Elasticities of Production and (33) Waugh, Frederick V. Demand and Price Analysis: Resource Earnings."J. Farm Econ. Vol. 49, No. 5, Some Examples from Agriculture. Tech. Bul. 1316, December 1967, pp. 1505-1510. Econ. Res. Serv., U.S. Dept. Agr., Washington, (22) Quance, Leroy. "Introduction To The Economic D.C., 1964. Research Service's Economic Projections Pro- (34) Yeh, Chung-Jeh. "Simulating Farm Outputs, gram." Paper prepared for 1975 Amer. Agr. Econ. Prices, and Incomes Under Alternative Sets of Assoc. annual meeting, Ohio St. Univ., Columbus, Demand and Supply Parameters." Paper prepared Ohio, Aug. 10-13, 1975. for 1975 Amer. Agr. Econ. Assoc. annual meeting, (23) Quance, C. Leroy and Luther Tweeten. "Excess Ohio St. Univ., Columbus, Ohio, Aug. 10-13, Capacity and Adjustment Potential in U.S. Agri- 1975.

64 •